Image registration for interventional MRI guided procedures: Interpolation methods, similarity measurements, and applications to the prostate

Baowei Fei, Zhenghong Lee, Jeffery L. Duerk, David L. Wilson

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

Nuclear medicine can detect and localize tumor in the prostate not reliably seen in MR. We are investigating methods to combine the advantages of SPECT with interventional MRI (iMRI) guided radiofrequency thermal ablation of the prostate. Our approach is to first register the low-resolution functional images with a high resolution MR volume. Then, by combining the high-resolution MR image with live-time iMRI acquisitions, we can, in turn, include the functional data and high-resolution anatomic information into the iMRI system for improved tumor targeting. In this study, we investigated registration methods for combining noisy, thick iMRI image slices with high-resolution MR volumes. We compared three similarity measures, i.e., normalized mutual information, mutual information, and correlation coefficient; and three interpolation methods, i.e., re-normalized sine, tri-linear, and nearest neighbor. Registration experiments showed that transverse slice images covering the prostate work best with a registration error of ≈ 0.5 mm as compared to our volume-to-volume registration that was previously shown to be quite accurate for these image pairs.

Original languageEnglish (US)
Pages (from-to)321-329
Number of pages9
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2717
StatePublished - Dec 1 2003
Externally publishedYes

Fingerprint

Image registration
Interpolation Method
Image Registration
Magnetic resonance imaging
Interpolation
Registration
High Resolution
Tumors
Mutual Information
Slice
Tumor
Nuclear medicine
SPECT
Ablation
Functional Data
Similarity Measure
Correlation coefficient
Medicine
Nearest Neighbor
Covering

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

@article{0a66834fd0d34de8b5a884bc8f9839ca,
title = "Image registration for interventional MRI guided procedures: Interpolation methods, similarity measurements, and applications to the prostate",
abstract = "Nuclear medicine can detect and localize tumor in the prostate not reliably seen in MR. We are investigating methods to combine the advantages of SPECT with interventional MRI (iMRI) guided radiofrequency thermal ablation of the prostate. Our approach is to first register the low-resolution functional images with a high resolution MR volume. Then, by combining the high-resolution MR image with live-time iMRI acquisitions, we can, in turn, include the functional data and high-resolution anatomic information into the iMRI system for improved tumor targeting. In this study, we investigated registration methods for combining noisy, thick iMRI image slices with high-resolution MR volumes. We compared three similarity measures, i.e., normalized mutual information, mutual information, and correlation coefficient; and three interpolation methods, i.e., re-normalized sine, tri-linear, and nearest neighbor. Registration experiments showed that transverse slice images covering the prostate work best with a registration error of ≈ 0.5 mm as compared to our volume-to-volume registration that was previously shown to be quite accurate for these image pairs.",
author = "Baowei Fei and Zhenghong Lee and Duerk, {Jeffery L.} and Wilson, {David L.}",
year = "2003",
month = "12",
day = "1",
language = "English (US)",
volume = "2717",
pages = "321--329",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Springer Verlag",

}

TY - JOUR

T1 - Image registration for interventional MRI guided procedures

T2 - Interpolation methods, similarity measurements, and applications to the prostate

AU - Fei, Baowei

AU - Lee, Zhenghong

AU - Duerk, Jeffery L.

AU - Wilson, David L.

PY - 2003/12/1

Y1 - 2003/12/1

N2 - Nuclear medicine can detect and localize tumor in the prostate not reliably seen in MR. We are investigating methods to combine the advantages of SPECT with interventional MRI (iMRI) guided radiofrequency thermal ablation of the prostate. Our approach is to first register the low-resolution functional images with a high resolution MR volume. Then, by combining the high-resolution MR image with live-time iMRI acquisitions, we can, in turn, include the functional data and high-resolution anatomic information into the iMRI system for improved tumor targeting. In this study, we investigated registration methods for combining noisy, thick iMRI image slices with high-resolution MR volumes. We compared three similarity measures, i.e., normalized mutual information, mutual information, and correlation coefficient; and three interpolation methods, i.e., re-normalized sine, tri-linear, and nearest neighbor. Registration experiments showed that transverse slice images covering the prostate work best with a registration error of ≈ 0.5 mm as compared to our volume-to-volume registration that was previously shown to be quite accurate for these image pairs.

AB - Nuclear medicine can detect and localize tumor in the prostate not reliably seen in MR. We are investigating methods to combine the advantages of SPECT with interventional MRI (iMRI) guided radiofrequency thermal ablation of the prostate. Our approach is to first register the low-resolution functional images with a high resolution MR volume. Then, by combining the high-resolution MR image with live-time iMRI acquisitions, we can, in turn, include the functional data and high-resolution anatomic information into the iMRI system for improved tumor targeting. In this study, we investigated registration methods for combining noisy, thick iMRI image slices with high-resolution MR volumes. We compared three similarity measures, i.e., normalized mutual information, mutual information, and correlation coefficient; and three interpolation methods, i.e., re-normalized sine, tri-linear, and nearest neighbor. Registration experiments showed that transverse slice images covering the prostate work best with a registration error of ≈ 0.5 mm as compared to our volume-to-volume registration that was previously shown to be quite accurate for these image pairs.

UR - http://www.scopus.com/inward/record.url?scp=0142152384&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0142152384&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0142152384

VL - 2717

SP - 321

EP - 329

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

ER -